There are approximately 3.7 million Americans with visual impairment, most of whom have central vision loss and this number is expected to rise as our population ages. Broadcast television, movies and the internet are major sources of information, independence and entertainment, but these media are difficult to use for people with central vision loss. There are few effective assistive devices for using these dynamic media and this is partly because we know relatively little about the perception of natural images viewed with non-foveal vision. This application examines how the visibility of information in natural images depends on contrast and on the presence of nearby features. We introduce new computational techniques that measure and manipulate the visibility of areas within real movies. Next, we use computational and behavioral techniques to segment image locations that are important for perception from areas that are unimportant. These approaches are brought together in a novel, low cost image enhancement system that will operate on standard computers. This system will be evaluated by a group of people with central vision loss and compared side-by- side with alternative methods. PUBLIC HEALTH RELEVANCE: Television, movies and the internet are a major source of information, independence and entertainment, but these media are difficult to use by people with central vision loss and there are few effective assistive devices available. This project contributes directly to public health by developing a low cost image enhancement system for use by this growing population of visually impaired people.